Pre-term R Introduction

library(spData)

## To access larger datasets in this package, install the spDataLarge
## package with: `install.packages('spDataLarge',
## repos='https://nowosad.github.io/drat/', type='source')`

library(sf)

## Linking to GEOS 3.9.0, GDAL 3.2.1, PROJ 7.2.1

library(tidyverse)

## -- Attaching packages --------------------------------------- tidyverse 1.3.0 --

## v ggplot2 3.3.3     v purrr   0.3.4
## v tibble  3.1.0     v dplyr   1.0.5
## v tidyr   1.1.3     v stringr 1.4.0
## v readr   1.4.0     v forcats 0.5.1

## -- Conflicts ------------------------------------------ tidyverse_conflicts() --
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

library(ggthemes)
library(ggspatial)

Loading data

The house dataset is part of the spData package. It contains data on 25,357 single family homes sold in Lucas County, Ohio between 1993 and 1998, based on data from the county auditor (this data is taken from the James P. LeSage’s Spatial Econometrics Toolbox for Matlab. The dataset includes the following variables:

• price: selling price
• yrbuilt: year built
• stories: one of seven values: one, bilevel, multilvl, one+half, two, two+half, or three
• TLA: total living area in square feet
• Wall: one of seven values: stucdrvt, ccbtile, metlvnyl, brick, stone, wood, partbrk
• beds: number of bedrooms
• baths number of full bathrooms
• halfbaths: number of half bathrooms
• frontage: frontage in feet
• depth: depth in feet
• garage: one of four values: no garage, basement, attached, detached, carport
• garagesqft: garage size in square feet, or zero for no garage
• rooms: number of rooms
• lotsize: lotsize in square feet
• sdate: sale date in format, yymmdd, e.g., Oct 17, 1997 = 971017
• avalue: auditor’s assessed value
• s1993: 1 indicated the home was sold in 1993, otherwise 0
• s1994: 1 indicated the home was sold in 1994, otherwise 0
• s1995: 1 indicated the home was sold in 1995, otherwise 0
• s1996: 1 indicated the home was sold in 1996, otherwise 0
• s1997: 1 indicated the home was sold in 1997, otherwise 0
• s1998: 1 indicated the home was sold in 1998, otherwise 0
• syear: Year the home was sold
• age: Age of the home in 1999, divided by 100

homes <- st_as_sf(house)

Scatter plots

The ggplot package is part of tidyverse, which was developed by Hadley Wickham. It offers a powerful set of tools for visualizing data, using an approach Wickham refers to as “a layered grammar of graphics,” which lets you create graphics using layers of commands.

Simple scatter plot

For the first “layer,” you’ll call the ggplot() function, which sets up the the plot and indicates that we’re working with the homes dataset we’ve loaded. Then you can add a layer to represent the data using geom_point(). You’ll need to specify which variable you’ll represent on the x-axis and which variable you’ll represent on the y-axis. Let’s make a quick scatterplot showing the year the home was built on the x-axis and the price of the home on the y-axis.

ggplot(homes) +
  geom_point(aes(x = yrbuilt, y = price))

In your own RStudio session, experiment with plotting other pairs of variables.

Additional variables

I can represent additional variables with color and size. For example, I might have different colors represent different types of buildings (the stories variable).

ggplot(homes) +
  geom_point(aes(x = yrbuilt, y = price, color = stories))

And I might use the sizes of the points to represent each home’s square footate (the TLA variable).

ggplot(homes) +
  geom_point(aes(x = yrbuilt, y = price, color = stories, size = TLA))

It’s sort of hard to see what’s going on with all those dots on top of each other, so you might want to make them all a little bit transparent using the alpha argument.

Note: Characteristics that represent variables should go inside the aes() function, and characteristics you want to apply to all variables should go outside the aes() function.

ggplot(homes) +
  geom_point(aes(x = yrbuilt, 
                 y = price, 
                 color = stories, 
                 size = TLA),
             alpha = 0.25)

In your own RStudio session, experiment with representing two or more of these variables in a variety of different ways. For if you want to go way beyond simple scatterplots, feel free to peruse the ggplot cheat sheet for inspiration.

Themes

You might also apply a theme to your scatterplot if you don’t like the default appearance. Here is the same scatterplot using a more minimalist theme.

ggplot(homes) +
  geom_point(aes(x = yrbuilt, 
                 y = price, 
                 color = stories, 
                 size = TLA),
             alpha = 0.25) +
  theme_bw()

And here it is using a theme inspired by plots that appear in the Wall Street Journal.

ggplot(homes) +
  geom_point(aes(x = yrbuilt, 
                 y = price, 
                 color = stories, 
                 size = TLA),
             alpha = 0.25) +
  theme_wsj()

Experiment with applying a few different themes to your scatterplot.

Map

A scatter plot is useful for showing how two or more variables relate to one another, but we may also be interested in how they vary across space. This dataset includes spatial information, so we can plot it on a map using geom_sf().

Let’s create a map showing how the price of a single-family home varies across space.

ggplot(homes) +
  geom_sf(aes(color = price), alpha = 0.5)

An appropriate theme for many maps is theme_map.

ggplot(homes) +
  geom_sf(aes(color = price), alpha = 0.5) + 
  theme_map()

You might want to orient your viewer using a basemap. The ggspatial package has a few you can choose from. Here’s the default.

ggplot(homes) +
  annotation_map_tile(zoomin = 0, progress = "none") +
  geom_sf(aes(color = price), alpha = 0.5) + 
  theme_map()

I also really like the black and white Stamen basemap.

ggplot(homes) +
  annotation_map_tile(zoomin = 0, progress = "none", type = "stamenbw") +
  geom_sf(aes(color = price), alpha = 0.5) + 
  theme_map()